Mesh collision avoidance

Abstract

The invention relates to a system (100) for segmenting an object in image data using model-based image segmentation, the system comprising a feature unit (120) for identifying features in the image data for computing an external energy of a mesh on the basis of a current position of the mesh, wherein the feature unit (120) further comprises a candidate feature unit (122) for selecting a plurality of candidate features in the image data, for identifying a feature to be included in the features identified in the image data, a position unit (124) for determining a position of each candidate feature of the plurality of the candidate features relative to a region of the image data, a feature function unit (126) for computing a strength of each candidate feature, wherein the strength of each candidate feature depends on the position of each candidate feature relative to the region, and an evaluation unit (128) for evaluating each candidate feature of the plurality of candidate features and for identifying the feature among the plurality of candidate features based on this evaluation. Determining whether a selected candidate feature is located inside the region which should be avoided, e.g., inside another mesh adapted to another object in the image data, allows penalizing this candidate feature during the computation of the strength of each feature and thus during the evaluation of the plurality of candidate features.

Description

FIELD OF THE INVENTION[0001]The invention relates to image segmentation and more particularly to simultaneous model-based segmentation of multiple objects in image data.BACKGROUND OF THE INVENTION[0002]Model-based segmentation has numerous applications in interventions and follow-up studies, for instance in radiation therapy planning. Deformable models, described by flexible meshes, for instance triangle meshes or simplex meshes, are adapted to the corresponding image structures. Usually this adaptation is carried out for every object separately by optimizing a weighted sum of two competing energies: an external energy driving the mesh triangles towards features in the image, and an internal energy preserving the form of the model. An implementation of this method is described by J. Weese, M. Kaus, C. Lorenz, S. Lobregt, R. Truyen, V. Pekar in “Shape constrained deformable models for 3D medical Image segmentation”, IPMI 2001, 3 pp. 80-387, hereinafter referred to as Ref 1.[0003]A se...

AI Technical Summary

Benefits of technology

[0005]It would be advantageous to provide means for segmentation of an object, while minimizing the likelihood of or preventing overlap of the adapted model mesh with a region of the image data.

[0011]Determining whether a selected candidate feature is located inside the region which should be avoided, e.g., inside another mesh adapted to another object in the image data, allows penalizing this candidate feature during the computation of the strength of each feature and thus during the evaluation of the plurality of candidate features. For example, this can be achieved by decreasing the feature strength such that the evaluation unit is more likely to select a candidate feature outside the region. Especially in areas where a lot of wrong image features similar to the correct image features occur, e.g., in the case of the vertebral column, the invention helps identifying the right image feature from the plurality of candidate features. However, if a prominent candidate feature of the image is found inside the region, it may still be taken into account. Advantageously, if the region is defined based on another mesh adapted to another object in the image, finding a prominent feature inside the other mesh may indicate that the other mesh is not well adapted to the other object. Employing an algorithm that iteratively adapts the other mesh to the other object in the image data while penalizing candidate features located inside the mesh, may result in that the other mesh is repelled from the region defined by the adapted mesh, thereby avoiding a collision (i.e. an overlap) of the two meshes.

[0017]In an embodiment of the system, the strength of each candidate feature inside the region is a function of the distance from each candidate feature inside the region to a boundary of the region. For example, the strength of the candidate feature may be a product, wherein a factor of this product depends on the distance from the candidate feature inside the region to a boundary of the region. This provides an easy and yet versatile way of taking into account effects of the position of the candidate feature relative to the region on the strength of the feature.

[0019]In an embodiment of the system, the strength of each candidate feature inside the region substantially decreases with the distance from each candidate feature inside the region to the boundary of the region. Candidate features deep inside the region are thus heavier penalized than candidate features close to the boundary of the region.

[0020]In an aspect of the invention, the system is used for segmenting a plurality of objects in image data by adapting a mesh to each object of the plurality of objects, wherein, during adaptation of a certain mesh to a certain object of the plurality of objects, the region is defined based on current positions of meshes adapted to objects of the plurality of objects other than said certain mesh. Thus, the system may be arranged to detect and possibly prevent overlap of multiple meshes used for segmenting a plurality of objects in the image data.

Problems solved by technology

A separate adaptation of multiple meshes cannot take spatial relations between several objects into account and hence often results in wrong adaptation results such as for instance intersecting meshes.

Images